Pork Barrel Presidente

8/4/2019 Pork Barrel Presidente

1/33

Presidential Pork Barrel Politics

Walter R. Mebane, Jr. Gregory J. Wawro

July 13, 2002

We thank Robert Houck and Daniel Kheel for assistance, and Sherri Wallace for helpful com-ments on earlier drafts. Houcks work was supported in part by Theodore J. Lowi, the John L.Senior Professor of American Institutions. Kheels work was supported by an endowment fromJonathan R. Meigs. Data were made available in part by the Cornell Institute for Social and Eco-nomic Research (CISER) and the Inter-University Consortium for Political and Social Research.Earlier versions of parts of this paper were presented at the 1993 Annual Meeting of the Midwest

Political Science Association, Palmer House Hilton, Chicago, IL, April 1517, and in seminars atthe University of North Carolina at Chapel Hill and at Princeton University. The authors bear soleresponsibility for all errors.

Associate Professor, Department of Government, Cornell University. 121 McGraw Hall, Ithaca,NY 148534601 (Phone: 607-255-2868; Fax: 607-255-4530; E-mail: [email protected]).

Assistant Professor, Department of Political Science, Columbia University. 741 Inter-national Affairs, New York, NY 10027 (Phone: 212/854-8540; Fax: 212/222-0598; E-mail:[email protected]).


2/33

Abstract

Presidential Pork Barrel Politics

Although the literature on pork barrel politics in the United States is decidedly

Congress-centered, we argue that presidents pursue strategies to target federal

expenditures to local areas to promote their goals of reelection and preservation of a policy

legacy. We develop and test a theory of expenditure targeting which delineates how

patterns of federal spending will vary depending on whether the spending is part of a

reward system for elites who have aided the presidents campaign efforts or is aimed

directly at attracting the support of ordinary voters. We also consider how the

institutional complexity of federal expenditures affects the ability of elites to manipulate

spending and the ability of voters to credit the president for such spending. An analysis of

data measuring the geographic distribution of federal spending during the second term of

the Reagan Administration provides support for our theory.


3/33

Forget the polls, I said. You cant beat an incumbent president. Re-

member, hes got a hundred billion dollars at his disposal to distribute to local

governments, and he can send that money anywhere he wants. Everybody from

Alabama to Alaska files for projects, and the administration decides which ones

to approve. In an election year, they go where the votes are.

Tip ONeill (ONeill and Novak 1987, 326)

The literature on the geographic distribution of federal government spending in the

United States is decidedly Congress-centered, practically ignoring the president (e.g., Fer-

ejohn 1974; Levitt and Snyder 1995; Stein and Bickers 1995; Bickers and Stein 2000; Lee

2000; two important exceptions are Mayer 1995 and McCarty 2000). It has become the

conventional wisdom that pork barrel politicsbringing home the baconplays a central

role in the electoral strategies of members of Congress (Mayhew 1974; Arnold 1979; Fiorina

1989; Levitt and Snyder 1997), but it is not known whether presidents pursue similar strate-

gies to deliver federal benefits in the form of local federal expenditures (LFEs) to electorally

relevant constituencies. While anecdotal evidence of presidents targeting LFEs for electoral

purposes abounds, systematic analysis regarding such actions is lacking.1

In this paper we examine whether patterns in the distribution of LFEs are consistent with

the presidents electoral interests and interests in preserving a policy legacy. We develop and

test a theory of expenditure targeting which delineates how patterns of federal spending

will vary depending on whether the spending is part of a reward system for elites who

have aided the presidents campaign efforts or is aimed directly at attracting the support

1For examples from recent presidential elections see James Gerstenzang, Santa Claus is an incumbent,states find as Bush dips into his federal goody bag, Los Angeles Times, 11 March 1992, 13A; Jack Andersonand Michael Binstein, Chowing down in Porkland, Washington Post, 24 Aug. 1992, 12C; Brian Kelly,Pigging out at the White House, Washington Post, 6 Sept. 1992, 1C; Jack Meyers and Ed Hayward, Clin-ton makes promi$e$ to Bay State, The Boston Herald, 29. Sept. 1996, 4; Brian McGrory, Clinton VowsMoney for Area Projects, The Boston Globe, 29 Sept. 1996, City Edition, Metro/Region, A1; David Dahl,How Bill Clinton Bought California, St. Petersburg Times, 15 Sept. 1996, South Pinellas Edition, Na-tional section, 1A ; Robert A. Rankin, Clinton gets creative with near-empty pork barrel, The Fort WorthStar-Telegram, 22 Sept. 1996, Final A.M. Edition, 26. See Mayer (1995, 167) for examples of announcementsof defense contract awards that were intended to influence the 1992 election.

1


4/33

of ordinary voters.2 Our theory predicts that after a successful reelection campaign, the

president rewards local elites who worked especially effectively in that effort by appointing

them to positions in his administration. These appointees become actors in policy subsystems

(Stein and Bickers 1995) where they can work with other elites back in the local area to

deliver their share of federal largesse. Thus in the first part of the presidents term (i.e.,

after the previous election), targeting ought to depend on the amount of support actually

received as newly appointed elites from areas that were important to the presidents election

direct spending back to the areas from whence they came. By midterm, however, most of

these appointees have left office (Heclo 1977, 103104). After the midterm election, the

president redirects some kinds of government spending to build support among voters for

the next election. Hence, in the second half of the presidents term (i.e, before the next

election) targeting ought to be based on the support expected from each area. We expect

this general argument to hold whether the president is in his first or second term, since a

lame-duck president has an interest in promoting the electoral chances of his partys successor

candidate in order to preserve the presidents policy legacy.

The types of spending that are targeted will vary depending on whether the spending

is elite-oriented or voter-oriented. A necessary condition for voters to respond to LFEs is

that the spending outcomes are traceable to presidential decisions (cf. Arnold 1990, 47).

Traceability would mean that the outcomes involve LFEs that each voter can both observe

and attribute to presidential actions. Elite responses may also require such causal and

retrospective analysis, but we posit that the connection to elites follows quite different kinds

of engagement focused on anticipated spending. The prospect of presidential patronage

should lead elites to expect that they will enjoy special advantages in competing for LFEs

when spending decisions are made.

Traceability is a function of the spendings institutional complexitythat is, the num-

ber of federal, state, and local government institutions that are involved in the expenditure of

2A key difference between our analysis and Mayers (1995) is that he examines electorally motivatedannouncements of expenditures, while we focus on actual expenditures.

2


5/33

federal funds. The institutional complexity of many kinds of LFEs ought to make it difficult

for anyone who is not immediately involved in the disbursement process to trace responsibil-

ity for the spending results back to the president. Among those not closely involved are most

voters. Spending that is primarily aimed to attract the support of voters directly should be

less institutionally complex than spending that is associated with elites support.

In our analysis we estimate targeting functions to test for the patterns posited by our

theory. We find that patterns in LFEs during the Reagan administration support our theory

in terms of the timing of elite-oriented and voter-oriented spending and the extent to which

expenditures are bound up with the institutions of American federalism.

The paper proceeds as follows. We first develop the theory of elite-oriented targeting,

detailing how presidents can use their appointment power to provide incentives for elites to

devote resources to the presidents campaign in exchange for influence over the distribution

of federal spending. We then discuss the issue of institutional complexity and how it affects

presidents targeting strategies. We then specify an econometric model of targeting and

report the results of the estimation of this model. We conclude with a discussion of the

implications of our findings for the literature on federal spending with respect to presidential,

congressional, and bureaucratic interactions.

Strategic Interactions with Local Elites: Delegating Patronage

Our theory begins with the fundamental assumption that the presidents main goal in tar-

geting LFEs is to get reelected, or in the case of a president in his second term, to aid the

election of his partys successor candidate, which is typically the vice president. Reelection

to a second term or the maintenance of party control of the White House promotes what

Moe and Howell (1999) contend is the presidents overarching goalthe establishment and

sustenance of a legacy.3

3We also assume the president is already committed to an overall election plan that includes such decisionsas which states to go after in the Electoral College (Brams 1978, 98107). Our concern is with the targetingstrategies the president may be expected to use in the context of such a plan.

3


6/33

The presidents strategy for targeting LFEs ought to depend on the effects he expects

spending in different geographic areas will have on his election prospects.4 The biggest

effects, and therefore the most vigorous targeting efforts, may or may not reflect direct

responses to LFEs by voters. If it is difficult for voters to trace responsibility for particular

expenditures back to the president, it may not pay the president to target based on their

anticipated direct reactions (cf. Stein and Bickers 1995). Plainly it is not easy for ordinary

voters to sort out exactly what the president has and has not done. Usually a multitude

of public officials and private groups claim credit for each governmental action or inaction.

The scope of federal spending is so vast that even the most dedicated voter would not be

able to comprehend it in detail in its entirety (cf. Bureau of the Census 198490). The fact

that a large amount of federal spending reaches voters indirectly further complicates the

tracing task. Substantial amounts of spending pass through state governments, for example,

often with each state supplementing the federal amount. Sometimes the level of federal

spending is pegged to the amount of state spending, as in the case of matching requirements

in grants for transportation construction. A voter who tried to measure the presidents

unique contribution across the full range of LFEs would face an impossible task.

In addition to the traceability problems, a president who tries to use LFEs directly to

attract voters may face a credibility problem. If voters act prospectively, then the president

has to convince voters that after the election he will deliver the amounts of LFEs they

prefer. Mere promising is likely not to be sufficient, for the challenger can make promises as

well (Downs 1957; Fiorina 1981). The only thing the president can do that the challenger

cannot do is to target some LFEs before the election, and to do so in a highly visible

way. Barons (1989) analysis of possible shirking by incumbents who receive campaign

contributions in anticipation of services they promise to provide after the election suggests

4Levitt and Snyder (1995, 961) contend that the Democratic party, despite its control of the House andthe Senate throughout most of the postwar period, has not been able to target extraordinary amounts offederal money to specific local areas, though they find that Democratic constituencies generally fare betterin the distribution of certain funds. Yet, their analysis does not take into consideration the presidentspreferences for LFEs, nor does their analysis include certain types of LFEs (e.g., military procurements andwages), which we suspect are ones that the president would be especially interested in targeting.

4


7/33

that the president will target some LFEs before the election, to signal voters regarding

the presidents commitment, even if the bulk of the targetingthe actual delivery of the

payoffs the president promisedtakes place afterward. Hence, even if voters are making

their decisions based solely on their expectations about future outcomes, the president needs

to act to appeal to their retrospective evaluations. Such a strategy will require targeting

actions late in the presidents term.

In light of the traceability problems, or perhaps simply in order to exploit all possible

electoral resources, the president may wish to use LFEs to appeal to the elites who have

interests in each local area and who control resources that can help or harm the presidents

reelection chances. It is such local elites who are involved in filing for projects (from

Alabama to Alaska, as Tip ONeill said) and in determining the local distribution of the

projects benefits and costs. Hence, at a minimum, local elites are more capable than ordinary

voters are of knowing both where money is being spent and where money is not being spent

when it might have been (Haider 1974). Local elites can more effectively trace responsibility

for particular spending decisions back to the president.

A presidential candidate who wishes to appeal to local elites has not only to answer

concerns elites may have about his credibility, but also needs to find a way to ensure that

the local elites in fact do work to advance his electoral goals. At least in the aggregate, a

candidate can verify directly how strongly voters in a locality have supported him simply

by counting the votes cast in his favor. But the effort local elites exert on his behalf is only

imperfectly observable.

One way to solve both the candidates credibility problem and the problem of verify-

ing local elites effort is for the candidate to commit to a plan to appoint specially chosen

local elites to positions in the administration in which they will have some discretion over

how LFEs are distributed geographically. We call this patronage delegation. That presi-

dents use appointments to deliver political patronage is a well established fact (Tolchin and

Tolchin 1971; Heclo 1977, 9299; Macy, Adams, and Walter 1983; Pfiffner 1988, 6889).

5


8/33

After the election, the White House is inundated with requests for appointments from those

who worked during the campaign (Macy et al. 1983, 4849). Among the roughly 4,000

civilian appointments a president typically makes, we distinguish two classes: partisan core

appointees, whose principal job is to advance the presidents ideological and policy agenda;

and local elite appointees, who do not oppose the presidents program but who are especially

concerned with returning federal benefits to their home localities. It is the latter class of

appointee that will most directly be involved in the delivery of LFEs.

Before each presidential election, the task of learning what actions will electorally most

benefit each candidate is given to the population of local elites who might wish to be ap-

pointed to federal offices if the candidate wins. The president can expect such would-be

appointees to be especially knowledgeable about the local areas with which they are most

strongly connected, and many will be strongly interested to be in a position to direct LFE

benefits to those areas. Local elites may have direct interests in the benefits, or they may

be interested in using the delivery of the LFEs to further their careers, political or other-

wise, by enhancing their local reputations or their connections with other local elites. Having

everyone understand that a suitable rule will be used to make appointments to the discretion-

carrying positions allows each candidate to give many local elites incentives to deploy their

political resources for the candidate in ways that are optimal for the candidate, even though

the candidate can only imperfectly observe local elites actions.

Local elites have recourse to two kinds of actions to enforce a candidates commitment to

a patronage delegation rule and hence make the commitment credible. Should the candidate

be elected and then renege, they may refuse to work for a possible reelection effort or for

the successor candidate of the same party. They may even refuse to work for the partys

candidates for other offices. The threat of such actions may not strongly constrain the

president, however, in part because the elites threat not to work in the future may not itself

be credible. But to the extent such threats are taken seriously, they may act even on a

lame-duck president based on the presidents commitment to his party and concerns about

6


9/33

preserving a policy legacy.

The other kind of action can affect the president more directly and immediately: local

elites may be able to use the connections they have with their congressional delegations

to interfere with appointments the president may attempt to make that deviate from the

generally understood appointment plan. Because the opposition of even a single senator

may suffice to derail a nomination, local elites who have the appropriate connections may be

able in effect to veto deviating appointments. Senators of the same party as the president

may be more susceptible to local elite threats not to work for the partys candidates than

the president is. Members of the House do not have the ability directly to intervene in the

confirmation process, though they may add their voices to the possible chorus of complaints

from local elites. But prompted by fire alarms (McCubbins and Schwartz 1984) sounded

by the local elites, Representatives may be able to use their oversight powers to make life

difficult for those appointed out of line with the patronage delegation norm. The anticipation

of such problems at least gives the president a strong incentive to consult with Congress about

appointments and hence may give local elites chances to enforce the patronage delegation

appointment rule.

Each candidate needs an appointment rule that encourages local elites to exert their

efforts in the most useful way during the campaign. The effort each local elite exerted

during the campaign is only imperfectly observable. So each candidate needs to choose a

rule that can be implemented using observable information and that leads each local elite to

act on the private information each has about his or her own capacity in the way that most

helps the candidates election chances.

We focus on a feasible rule that we describe as converted-loss. The rule aims to motivate

local elites to work where they think their effort will convert a probable loss into a probable

win. The rule specifies that the highest probability of getting an appointment will go to

those who convert a pre-election expectation that the candidates share of the two-party

vote will be less than .5 into a election-day result of greater than .5. To observe increases in

7


10/33

support, each candidate must be able to measure the level of support he has in each local

area before the campaign period begins. It is reasonable to assume that focused opinion

polling and other sources give the major candidates fairly accurate information about such

levels.

A downside of such a converted-loss approach is that the president cannot be confident

that the local elites he appoints will share his policy goals. Voting in presidential elections

is strongly partisan. Elites who would be most effective in mobilizing votes in a candidates

geographic areas of strongest support are likely to endorse at least the broad, partisan

outlines of the candidates policy preferences. But local areas that can be converted are

likely to be areas in which many voters do not identify with either party and are only weakly

committed to the core policy stands of any candidate. The elites who can best swing voters

in such areas to support a particular candidate may not be the elites most strongly aligned

with that candidates policy commitments.

Appropriate procedures for screening and for defining appointees authority can substan-

tially reduce this defect in the converted-loss approach.5 Calvert, McCubbins, and Weingast

(1989) show that under a wide variety of conditions an appointees decisions will be con-

strained to lie in a region bounded by the presidents and the Congresss preferred spending

positions.6 So local elites ought not to exert campaign effort for a candidate whose policy

goalsespecially with respect to preferences for spendingare very different from their own

if they expect that candidate to use effective screening and review procedures. Seriously

discrepant local elites ought to anticipate that they will not receive appointments, or that

the presidents central monitoring agent (e.g., the Office of Management and Budget [OMB])

will veto their efforts to implement their preferred spending patterns. A converted-loss ap-

proach coupled with screening and review can therefore give the president appointees who

5Pfiffner (1988, 6869) reviews the appointments procedures used in presidential administrations andnotes the especially rigorous screening procedures used by the Reagan administration and the first Bushadministration.

6The president has institutional advantages that enable him to appoint individuals to government posi-tions that are more sympathetic to his policy goals than they are to those of members of Congress (Moe1987). Thus appointees decisions may be skewed toward the presidents preferred spending position.

8


11/33

worked as effectively as they could to support his election, and who support his preferences

for the distribution of LFEs as closely as possible given the preferences and powers of the

Congress.7

But if each appointee has information superior to the presidents (and the Congresss)

about the capacity or tastes for LFEs in at least one local area, then there will be an ele-

ment of uncertainty that increases each appointees discretion in making spending decisions

(Calvert et al. 1989, 597). Local elites who secure appointments can use this discretion

to extract the rewards that presumably helped to motivate them to work for the winning

candidate in the first placefor example, extra LFEs being directed to the local areas the

elites are associated with. Being a patron-delegate may be a good way to ensure that one

receives ones share of direct patronage. If an appointed local elite has only a small infor-

mational advantage over the presidents central monitor (such as OMBsee Pfiffner 1979),

then the amount targeted through such discretion will be small relative to the total amount

of expenditure over which the elite has authority (cf. Chubb 1985).

Targeting and Institutional Complexity

LFEs can help attract active support from local elites. They may also help win votes directly

if voters like the patterns of expenditure they see and can trace responsibility for them back

to the president. The president will face a dilemma if the strategy that best appeals to

voters is incompatible with the one best suited to local elites. This should be true if, as we

expect, appointment-seeking local elites care about the absolute levels of their post-election

7The converted-loss appointments strategy can maximize the effort locally interested elites exert on thepresidents behalf, and pre-appointment screening can minimize the chances that an appointee has policygoals at wide variance from the presidents. But a presidential candidate would gain electorally from using

the converted-loss strategy even if he or she were completely indifferent to the geographic distribution ofLFEs. Indeed, such indifference can attract effort in favor of the candidate from the widest range of localelites, if local elites can reasonably expect to encounter minimal screening of their geographic preferences.In this case the president would be largely uninvolved in the post-election efforts to target spending, andspending outcomes would be determined through interactions between the bureaucracy and Congress. Evenin this case, however, bureaucratic appointees preferences would be shaped by the processes that selectedthem, so that the interactions with Congress would be strongly affected by the geographic distribution ofelectoral support for the president

9


12/33

LFE rewards, while retrospective voters who want to know what have you done for me

lately care about recent changes in LFEs. Maximizing post-election levels for local elites

may make it difficult to produce the pre-election changes that voters would like. The need

to choose between local elites and voters may be obviated, however, if elites and voters do

not care equally intensely about the same kinds of LFEs. On the one hand, LFEs that elite-

appointees can control relatively easily but that are virtually invisible or unfathomable to

voters ought to be prime territory for elite-oriented maneuvers. On the other hand, features

of LFEs that make them easy for voters to trace to the president may render them less

attractive to local elites.

What we call the institutional complexity of LFEs is likely to affect both their susceptibil-

ity to local elite control and the difficulty voters face in tracing them. The most complex are

LFEs that pass through state or local governments before being disbursed into private hands,

and over which the lower-level governments exercise some autonomous authority. Such LFEs

offer local elites many opportunities to influence who receives money, because institutional

complexity creates many decision points at which they may intervene. Elite-appointees are

often closely connected to state and local governments and parties (e.g. Mackenzie 1981, 64).

Intergovernmental LFEs often closely match local needs (Peterson, Rabe, and Wong 1986,

8194). We would expect that appointees often coordinate with officials at several levels

of government to deliver special intergovernmental LFEs to their home areas (cf. Haider

1974). Anyone not familiar with the negotiations that produce such expendituressuch as

ordinary votersmay have a hard time deciding whom to credit for the results.

LFEs that go directly from the Federal government into individuals hands are less com-

plex and are the easiest for individual voters to trace. Examples of such LFEs include military

and civilian employment and procurements, and transfer payments to individuals for social

security retirement pensions, and other direct Federal payments. Some administrations have

gone to great lengths to make sure recipients of these kinds of spending credit the president

for their good fortune (Tufte 1978, 3032). Education, highway, and social welfare transfers

10


13/33

are not perfectly traceable to the Federal level, because states and localities help fund some

kinds of payments and expenditures and have authority over some payment levels (Peterson

and Rom 1990). But to a great extent transfer payments depend on entitlement rules and

mandates set at the Federal level.

The targeting strategy to be expected for local elites is the one induced by the converted-

loss appointments strategy. If the president is rewarding local elites through such a strategy,

then after the election LFE levels ought to be highest in those areas where the president

received more than half the votes, but less than an overwhelming proportion. Recall that

a converted-loss strategy would make elites from such areas the most likely to receive new

appointments and therefore put them in position to direct LFEsnew government jobs,

procurements, intergovernmental grantsback to those areas.

One might well say that, strictly speaking, the president does not have a targeting strat-

egy with respect to local elites. Rather we expect the actions of elite-appointees to induce a

pattern in which there is more spending in areas where the president received somewhere in

the range of 5070% of the votes than there is in other areas. Patronage delegation ought

to produce such elite-induced targeting. Yet since the discretionary authority of local elites

newly appointed after the election lasts only as long as they hold office, which is typically

less than two years (Heclo 1977, 103), we expect this pattern to dominate only during the

first part of the presidents term.8

We expect to see more voter-oriented targeting in the second-half of the term, which

means that the type of LFEs involved should be less institutionally complex than those

targeted in the first-half. Yet at least in terms of the relation between LFEs and expectations

of voter support, similar types of geographic areas ought to be targeted in both cases. Under

reasonable assumptions about the distribution of voter support (such as that individual

voters each choose according to a probit model), the geographic areas where LFEs are most

likely to convert losses to wins are the areas where the marginal gain from each unit of

8Those who were appointed after the election are usually replaced with more career-oriented civil servantswho would have much weaker ties to local areas throughout the country.

11


14/33

spending is the largest (Wright 1974; cf. Dixit and Londregan 1996). The scope and sizes of

the changes the president can produce in LFEs to appeal directly to voters are not unlimited.

The president will want to deploy the changes he can make to produce the largest possible

increase in his reelection prospects. This implies, similar to the elite-oriented strategy, that

such changes will be concentrated in the areas where the president expects they are most

likely to convert a loss to a win. Relative to the pre-election information the president uses

to make the changes, we ought to observe the largest changes where his prospects look bad

but not hopelesssay areas where his pre-election support appears to be somewhere in the

range of 3050%.9

An Statistical Model for Targeting and Voting

To summarize our theoretical development, targeting can be elite-oriented or voter-oriented.

Elite-oriented targeting is induced by appointees who are selected based on the support

the president received in the most recent presidential election. Voter-oriented targeting is

based on expectations of support in the upcoming election for the president (or his partys

successor candidate) based on information obtained during the presidents term. Elite-

oriented targeting should occur in the first part of the presidents term, as elite-appointeesreap the rewards of helping the president gain office. Since appointees on average serve less

than two years, voter-oriented targeting should occur in the second-half of the presidents

term. The politics of the midterm election should also inform assessments of support for the

president in different local areas. The types of LFEs involved in targeting should vary in

terms of their institutional complexity across the pre- and post-midterm periods.

The president cares about the aggregate distribution of the vote in each area. Suppose

each individual voter i in local area s in year t decides whether to vote for the incumbent

based on a continuous index yist, according to the rule yist = 1 if yist > 0, yist = 0 if

yist 0, where yist = 1 indicates a vote for the incumbent and yist = 0 indicates a vote

9When targeting voters, presidents can rely on institutional resources, such as the veto (McCarty 2000)or OMB monitoring, to direct spending to particular constituencies.

12


15/33

against the incumbent. Given the Electoral College, the presidents direct interest is in

winning half or more of the electoral votes. Because a direct model of the presidents choice

among all the possible electoral vote majority patterns would be unwieldy, we simplify by

assuming that the president is interested in the aggregate vote propensity in each local area.

Specifically, we assume that the president cares about the mean value of yist in area s, i.e.,

about yst = N1st

iVst

yist, where Vst denotes the set of voters in area s at time t and Nst

denotes the number of such voters.

The administration targets LFEs to each area during each year t to match his support

there, according to a particular strategy. The strategy is a function of the information yst

that the president has at time t about yst, and is tailored to each type of expenditure. We

use Sjk to denote the targeting function the president uses for the kth type of LFE during

either the early period j = E or the later period j = L. For each value of yst, this function

indicates how much the supply of the kth type of LFE to area s during t ought to deviate

from a baseline value. The set of possible strategies for each kind of LFE includes the null

possibility that the president does no targeting related to yst at all. In this case Sjk(yst) = 0.

Consistent with our theory, we allow the targeting function to differ between the first

two years and the last two years of the presidents term; that is, possibly SEk = SLk. Let

ht = 0 during the first two years and ht = 1 during the second two years, and let zst be

a fixed vector of observed exogenous variables, k0 be a scalar and k a vector of constant

coefficients, and ks and kt be respectively area-specific and time-specific fixed effects. Our

specification for the amount xkst of the kth type of LFE going to area s in year t is

xkst = exp[k0 + (1 ht)SEk(yst) + htSLk(y

st) + z

stk + ks + kt] + ukst (1)

The area-specific effects would capture any adjustment in the level of the LFE done through-

out whole states pursuant to a plan to build a majority in the Electoral College. The distur-

bance ukst has expectation E[ukst] = 0 and variance E[u2

kst] = E[xkst] 2

k and is uncorrelated

13


16/33

across k, s, and t. We use the exponential form in equation (1) because virtually all observed

local aggregations of LFEs are nonnegative. An additive disturbance with the specified form

of heteroscedasticity is frequently appropriate for models with log-linear expectations (Mc-

Cullagh and Nelder 1989, 193ff).

Targeting Model Specifications and Results

To estimate the relationships between LFE levels and support across geographic areas, we

use annual, county-level data for years 198588 to estimate equation (1) separately for each

of sixteen kinds of LFEs.10 Analyzing these years provides an interesting test of our theory

since, with the possible exception of defense spending, the Reagan administration opposed

government spendingat least in rhetoric. It also enables us to test the application of our

theory to a lame-duck president. Although the incumbent president did not run in 1988, we

expect our theory to hold because the Reagan administration went to great lengths to assist

George Bush, Sr. in his friendly takeover of the White House. Reagan traveled extensively

to campaign for Bush and personnel changes in the cabinet in 1988 were molded to fit Bush

priorities (Pfiffner 1990, 65).

The specification is designed to check whether LFE targeting behavior varies during thepresidents term. We focus on the distinction between targeting early and targeting late

in the term, using the midterm election as the dividing point. To measure the presidents

information yst we use two different quantities. During the first two years of the term,

we measure information by the proportion of the two-party vote cast for the incumbent in

each county in the preceding presidential election, transformed using the inverse Normal

cumulative distribution function. That is, during the first two years, yst = 1(psP), where

psP denotes the preceding presidential election vote proportion in county s. Since we are not

privy to internal polling the president does to assess his support in different areas, during

the second two years of the term we use NES data to simulate the support the president (or

10The counties in the analysis are 138 of the 139 counties included in the 1986 NES sample (one countywas omitted due to missing data), which we use to compute midterm support for the president.

14


17/33

his successor) might expect to receive at midterm from each county if individual voters were

to use at that time the same voting decision rule they used in the preceding election. This

approach gives post-midterm information values computed using yst = 1(psM), with

psM = n1sM

ivsM

pisM (2)

where pisM is the predicted probability that voter i surveyed in the midterm NES would

vote for the president, and nsM is the number and vsM is the set of all voters in the midterm

sample in county s. Details regarding the computation of the probabilities pisM appear in

the Appendix.

We use polynomials to approximate the targeting functionsSEk and

SLk. For each kind

of LFE k we define

SEk(yst) =

mk

r=1

kr(yst)

r (3a)

SLk(yst) =

mk

r=1

kr(yst)

r . (3b)

for constant coefficients kr and kr and some choice of polynomial degree mk. To choose mk

we use F-tests based on deviance statistics.

The variables zkst include a set of dummy variables that represent the annual member-

ships on fifteen standing committees of the U.S. House delegation from the state to which

county s belongs, and a variable that measures the annual average for the state of the num-

ber of consecutive terms served by the states House members.11 Admittedly, these variables

are crude measures of congressional influence on LFEs, but should serve our purpose since

11Each committee dummy variable equals one if any member of the delegation is a member of the com-mittee in the referent year; otherwise the variable equals zero. We used dummy variables for the followingcommittees: Agriculture; Appropriations; Armed Services; Banking, Finance and Urban Affairs; Budget;Education and Labor; Energy and Commerce; Government Operations; Interior and Insular Affairs; Mer-chant Marine and Fisheries; Post Office and Civil Service; Public Works and Transportation; Science, Spaceand Technology; Veterans Affairs; Ways and Means. Committee data were compiled from Joint Committeeon Printing (198387). Terms data were provided by Robert Stein.

15


18/33

our focus is on the president.12

The sixteen LFEs cover most federal expenditures that disburse funds in particular ge-

ographic areas in the United States. The LFE variables include transfer payments, civilian

and military employment, salaries and procurements, direct payments (not for individuals

and not procurements), and intergovernmental transfers to local governments for education,

for highways, for social welfare or for other purposes. The LFE variables are listed, with

source citations, in Table 1. For the LFE variables that measure money the values are the

total dollar amount spent in each county in each year, divided by the county population

in that year; units are $1000 per person. For the variables that measure employment, the

values are the total number of jobs, per capita, with units being jobs per person. In light

of the fact that many Federal grants that are ultimately disbursed by local governments are

passed through state governments, we analyze, separately, transfers to local governments

from state governments as well as transfers from the Federal government.

*** Table 1 about here ***

The estimated targeting function polynomials are plotted in Figures 1 and 2.13 The

values represent partial LFE levels (net of congressional influences and fixed effects), plotted

against the support scores psP and psM. The ranges of values for psP and psM used in the

figures are the values that occur for the counties in the NES data. Figure 1 shows the pre-

midterm polynomials and Figure 2 shows the post-midterm polynomials. Table 2 shows the

point estimate and 95% confidence interval for the local maximum of each curve on the [0, 1]

interval, when such exists.

*** Figure 1, Figure 2, and Table 2 about here ***

The estimates suggest that targeting was induced by patronage delegation for most kinds

of LFEs in the pre-midterm period. Recall that for elite-induced targeting we expect to see

the most spending in local areas where the president received about 5070% of the votes. The

12The area-specific effect in equation (1) should also help account for congressional influences.13Estimates were computed using quasi-likelihood methods, using the glm function of S-PLUS (Statistical

Sciences 1991).

16


19/33

figures and Table 2 show the expected pattern during the pre-midterm period for eleven kinds

of LFEs. For two of those LFEsmilitary procurements and Federal highways transfers

the local maximum of the polynomial is not determined very precisely; the 95% confidence

intervals are wide. So perhaps they should not be counted as strongly matching the elite-

oriented pattern. The other five LFEs do not show the expected pre-midterm pattern.

Whether one counts the number of LFEs that match the predicted pattern as eleven or nine,

the number is substantially larger than the number of LFEs that exhibit the elite-oriented

pattern after the midterm election. During the post-midterm period only one LFEtransfer

paymentscomes close to showing the pattern, and that one estimate is not a strong match:

while the point estimate for the polynomials local maximum (.74) is not far above the upper

end of the elite-oriented range, the 95% confidence interval covers virtually the entire unit

interval.

To check for voter-oriented targeting we need to measure the changes that occur over

time in each kind of LFE as a function of the presidents support in each local area. We use

the targeting polynomial estimates for this purpose by computing the difference between the

predicted post-midterm and pre-midterm levels of each LFE in an idealized circumstance

where the presidents support remains constant at some level throughout the term. That

is, we compute the difference SLk(y) SEk(y

) for a range of y values. For voter-oriented

targeting we expect to see local maxima of the simulated changes in spending occur for

support values between roughly .3 and .5.

The simulated differences plotted in Figure 3 and the 95% confidence intervals for local

maxima shown in Table 3 suggest that voter-oriented targeting occurred for almost all of the

LFEs we rated as relatively low in institutional complexity but occurred for at most half of

the LFEs we rated as high in institutional complexity. Among the institutionally less complex

LFEs, only civilian employment does not have a local maximum in the voter-oriented range.

For military procurements, military salaries and transfer payments the local maxima are

global maxima. Among institutionally complex LFEs the voter-oriented pattern occurs for

17


20/33

three of the four Federal transfers. The pattern is only weakly matched for Federal highways

transfers, for which the 95% confidence interval spans virtually the entire unit interval.

For state highways transfers both the point estimate and 95% confidence interval of the

local maximum take values that seem a bit too low to support a voter-oriented targeting

interpretation.

*** Figure 3 and Table 3 about here ***

Conclusion

The distribution of LFEs is consistent with the presidents electoral interests, and generally,

his interests in preserving a policy legacy through the support of the electoral interests of

his partys successor candidate. We find evidence of both voter-oriented and elite-induced

targeting of LFEs for electoral purposes. Voter-oriented targeting occurs more often for LFEs

that are relatively low in institutional complexity and occurs later in the presidents term.

Voters can much more readily trace such LFEs back to the president than they can LFEs

that are institutionally complex. But the elite-induced targeting pattern occurs for both not-

so-complex and highly complex LFEs, including six of the eight kinds of intergovernmental

transfers. Targeting in these cases follows the pre-midterm pattern to be expected if eliteswho are interested in returning benefits to their local areas in the form of LFEs are appointed

to the administration according to the converted-loss strategy.

Our findings imply an important correction to work that has demonstrated how strongly

members of Congress can influence bureaucratic decisions regarding the geographic alloca-

tion of federal expenditures. Any theory of influence that neglects the presidents role and

interests in determining geographic spending outcomes is incomplete.14 But our results do

not necessarily contradict existing Congress-centered analyses because the patterns we es-

timate are net of congressional influences. One possible extension of the current analysis

is to examine more closely the conflicts and complementarities between the LFE targeting

14Granted, some Congress-centered studies of the distribution of LFEs acknowledge the limited scope oftheir analyses (e.g., see Arnold 1979, 19).

18


21/33

strategies of the president and members of Congress.

We have not emphasized some issues that have received a good deal of attention in dis-

cussions of pork barrel politics in Congress, most notably the role of organized interests

and the potential importance of logrolling (Ferejohn 1974; Shepsle and Weingast 1981; Stein

and Bickers 1995). We assume that organized interests are important in many ways in in-

teractions between the president and local elites. Logrolling is not a concern as long as

expenditures result from discretionary actions by appointed elites who act in isolation and

without substantial constraints. But it becomes more important when officials or agencies

must coordinatewith one another or with Congressto produce expenditures, or when

spending is subject to budget constraints. Nonetheless, it is reasonable to speak ofpresiden-

tial pork barrel politics. The president is often viewed as a national leader who represents

the entire country and is far less concerned than are members of Congress with local matters

such as where federal dollars get spent. However, the analysis in this paper indicates that

presidential politics surrounding federal expenditures, like congressional politics surrounding

such expenditures, is substantially local.

Appendix

To produce the probabilities pisM used in equation (2), we estimate an individual-level probit

regression equation using NES data from the preceding presidential election, at time P =

1984 (Miller and the National Election Studies 1986), based on yisP = a0 + wisPa + isP,

where wisP denotes a vector of survey opinion variables and the disturbance is assumed to

satisfy isP N(0, 1). To measure vote choices we used self-reports (1984 NES var. 787, 788),

omitting those validated not to have voted (var. 1130). wisP contains seven variables. Four

variables count the number of mentions for each of the open-ended party likes and dislikes

items: number of likes about the Democratic party (var. 266271); number of dislikes about

the Democratic party (var. 272277); number of likes about the Republican party (var. 278

19


22/33

283); number of dislikes about the Republican party (var. 284289). Two variables measure

retrospective economic evaluations, referring respectively to the nations economy over the

past year (var. 227) and to family finances now compared to a year ago (var. 139); values

of DK or NA are treated as missing. The seventh variable is the feeling thermometer for

Republican presidential candidate Reagan (var. 290); values of NA are treated as missing but

responses of doesnt recognize name or dont know where to rate are assigned the middle

value of 50. An observation missing data on any variable is omitted from the analysis. We

use the estimated parameters with data for the same variables from the midterm (M = 1986)

NES to compute pisM = (a0 + wisMa). The variables from the 1986 NES data are var.

7295 (likes and dislikes), var. 372, 355 (economic evaluations) and var. 130 (Reagan feeling

thermometer) (Miller and the National Election Studies 1987).

20


23/33

References

Arnold, R. Douglas. 1979. Congress and the Bureaucracy: A Theory of Influence. New

Haven: Yale University Press.

Arnold, R. Douglas. 1990. The Logic of Congressional Action. New Haven: Yale University

Press.

Baron, David P. 1989. Service-induced Campaign Contributions, Incumbent Shirking, and

Reelection Opportunities. In Peter C. Ordeshook, editor, Models of Strategic Choice in

Politics, Ann Arbor: University of Michigan Press.

Bickers, Kenneth N. and Robert M. Stein. 2000. The Congressional Pork Barrel in aRepublican Era. Journal of Politics 62: 10701086.

Brams, Steven J. 1978. The Presidential Election Game . New Haven: Yale University Press.

Bureau of Economic Analysis. 1990. Personal Income by Major Source and Earnings by

Major Industry, Counties, 19691988.. Computer data file. Washington, DC: U.S. Bureau

of Economic Analysis.

Bureau of the Census. 198490. Consolidated Federal Funds Report, Fiscal Years 1983-

88.. Computer data files. Washington DC: U.S. Department of Commerce, Bureau of the

Census. Albany, NY: State Data Center Program (distributor, 1984). Ann Arbor: Inter-

university Consortium for Political and Social Research (distributor, 1983, 198590).

Bureau of the Census. 198691. Annual Survey of Governments, Finance Statistics, 1984

88.. Computer data files. Washington, DC: U.S. Department of Commerce, Bureau of

the Census.

Bureau of the Census. 1991. Census of Governments 1987, Governmental Finances.. Com-

puter data file. Washington, DC: U.S. Department of Commerce, Bureau of the Census.

21


24/33

Calvert, Randall L., Mathew D. McCubbins, and Barry R. Weingast. 1989. A Theory

of Political Control and Agency Discretion. American Journal of Political Science 33:

588611.

Chubb, John E. 1985. The Political Economy of Federalism. American Political ScienceReview 79: 9941015.

Dixit, Avinash and John Londregan. 1996. The Determinants of Success of Special Interests

in Redistributive Politics. Journal of Politics 58: 11321155.

Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper & Row.

Ferejohn, John A. 1974. Pork Barrel Politics: Rivers and Harbors Legislation, 19471968.

Stanford: Stanford University Press.

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections . New Haven:

Yale University Press.

Fiorina, Morris P. 1989. Congress: Keystone of the Washington Establishment. New Haven:

Yale University Press 2 edition.

Haider, Donald H. 1974. When Governments Come to Washington: Governors, Mayors and

Intergovernmental Lobbying. New York: Free Press.

Heclo, Hugh. 1977. A Government of Strangers . Washington, DC: Brookings.

Joint Committee on Printing. 198387. Official Congressional Directory. Washington, DC:

U.S. Government Printing Office.

Lee, Francis E. 2000. Senate Representation and Coalition Building in Distributive Politics.

American Political Science Review 94: 5972.

Levitt, Steven D. and James M. Snyder, Jr. 1995. Political Parties and the Distribution of

Federal Outlays. American Journal of Political Science 39: 958980.

22


25/33

Levitt, Steven D. and James M. Snyder, Jr. 1997. The Impact of Federal Spending on

House Election Outcomes. Journal of Political Economy 105: 3053.

Mackenzie, Calvin. 1981. The Politics of Presidential Appointments . New York: Free Press.

Macy, John W., Bruce Adams, and J. Jackson Walter, editors. 1983. Americas Unelected

Government: Appointing the Presidents Team. Cambridge: Ballinger.

Mayer, Kenneth R. 1995. Electoral Cycles in Federal Government Prime Contract Awards:

State-Level Evidence from the 1988 and 1992 Presidential Elections. American Journal

of Political Science 39: 162185.

Mayhew, David R. 1974. Congress: The Electoral Connection. New Haven: Yale University

Press.

McCarty, Nolan. 2000. Presidential Pork: Executive Veto Power and Distributive Politics.

American Political Science Review 94: 117129.

McCubbins, Mathew D. and Thomas Schwartz. 1984. Congressional Oversight Overlooked:

Police Patrols versus Fire Alarms. American Journal of Political Science 28: 165179.

McCullagh, Peter and J. A. Nelder. 1989. Generalized Linear Models . London: Chapman

and Hall 2 edition.

Mebane, Walter R., Jr. 1993. Analyzing the Effects of Local Government Fiscal Activity I:

Sampling Model and Basic Econometrics. Political Analysis 4: 136.

Miller, Warren E. and the National Election Studies. 1986. American National Election

Study, 1984: Pre- and Post-Election Survey.. Computer file. ICPSR 8298. Ann Arbor, MI:

Center for Political Studies, University of Michigan. [original producer]. 2nd ICPSR ed.

Ann Arbor, MI: Inter-university Consortium for Political and Social Research [producer

and distributor].

23


26/33

Miller, Warren E. and the National Election Studies. 1987. American National Election

Study, 1986.. [computer file]. Ann Arbor, MI: Center for Political Studies, University of

Michigan. [original producer]. 2nd ICPSR ed. Ann Arbor, MI: Inter-university Consortium

for Political and Social Research [producer and distributor].

Moe, Terry M. 1987. An Assessment of the Positive Theory of Congressional Dominance.

Legislative Studies Quarterly 11: 475521.

Moe, Terry M. and William G. Howell. 1999. Unilateral Action and Presidential Power: A

Theory. Presidential Studies Quarterly 29: 850872.

ONeill, Thomas P. and William Novak. 1987. Man of the House: The Life and Political

Memoirs of Speaker Tip ONeill. New York: Random House.

Peterson, Paul E., Barry G. Rabe, and Kenneth K. Wong. 1986. When Federalism Works .

Washington, DC: Brookings.

Peterson, Paul E. and Mark C. Rom. 1990. Welfare Magnets: A New Case for a National

Standard. Washington, DC: Brookings.

Pfiffner, James P. 1979. The President, the Budget, and Congress: Impoundment and the1974 Budget Act. Boulder: Westview Press.

Pfiffner, James P. 1988. The Strategic Presidency: Hitting the Ground Running. Chicago:

Dorsey Press.

Pfiffner, James P. 1990. Establishing the Bush Presidency. Public Administration Review

50: 6473.

Shepsle, Kenneth and Barry Weingast. 1981. Political Preferences for the Pork Barrel: A

Generalization. American Journal of Political Science 25: 96111.

Statistical Sciences. 1991. S-PLUS.. Computer software. Version 3.0. Seattle: Statistical

Sciences, Inc.

24


27/33

Stein, Robert M. and Kenneth N. Bickers. 1995. Perpetuating the Pork Barrel: Policy

Subsystems and American Democracy. Cambridge: Cambridge University Press.

Tolchin, Martin and Susan Tolchin. 1971. To the Victor . . . : Political Patronage from the

Clubhouse to the White House . New York: Random House.

Tufte, Edward R. 1978. Political Control of the Economy. Princeton: Princeton University

Press.

Wright, Gavin. 1974. The Political Economy of New Deal Spending: An Econometric

Analysis. Review of Economics and Statistics 56: 3038.

25


28/33

Table 1: Types of local federal expenditure

InstitutionalVariablea Description complexity

transfer paymentsb,e transfer payments to individuals lowcivilian employmentb,f Federal government civilian employment lowmilitary employmentb,f Federal government military employment lowcivilian salariesc,e salaries and wages, all civilian and Postal Service low

employeesmilitary salariesc,e salaries and wages, all military personnel lowcivilian procurementsc,e procurement contracts, all except Defense low

Departmentmilitary procurementsc,e procurement contracts, Defense Department lowdirect paymentsc,e direct payments other than for individuals loweducation transfersd,e transfers to local governments for education highhighways transfersd,e transfers to local governments for highways highsocial welfare transfersd,e transfers to local governments for public welfare, high

employment security, health and hospitals, hous-ing

other transfersd,e

all other transfers to local governments high

Notes:a All variables are used per capita, based on county populationb.b source, Bureau of Economic Analysis (1990).c source, Bureau of the Census (198490).d source, Bureau of the Census (198691) and Bureau of the Census (1991); county totalsare estimated as in Mebane (1993).e units, $1000 per person.f units, jobs per person.


29/33

Table 2: Point Estimates and 95% Confidence Intervals of Support Values in the OpenInterval (0,1) That Maximize Local Federal Expenditures

Pre-Midterm

Maximum is an elite-oriented targeting value Maximum is not in the elite-oriented range

civilian procurements .61 (.61, .61) civilian employment .96 ( .00, 1.00)military procurements .68 (.18, .97) military salaries no maxa

military employment .71 (.71, .71) transfer payments no maxa

civilian salaries .68 (.67, .68) Federal other transfers .18 (.02, .56)direct payments .52 (.46, .58) State welfare transfers .29 (.24, .35)Federal highways transfers .67 (.00, 1.00)Federal welfare transfers .64 (.63, .65)Federal education transfers .56 (.52, .59)State highways transfers .67 (.67, .67)

State education transfers .55 (.53, .57)State other transfers .71 (.68, .74)

Post-midtermMaximum is an elite-oriented targeting value Maximum is not in the elite-oriented range

transfer payments .74 (.00, 1.00) civilian procurements no maxa

military procurements .44 (.32, .57)civilian employment .85 (.00, 1.00)military employment .04 (.02, .06)civilian salaries no maxa

military salaries no maxa

direct payments .90 (.43, 1.00)Federal highways transfers .98 (.00, 1.00)Federal welfare transfers .90 (.89, .90)Federal education transfers .99 (.83, 1.00)Federal other transfers no maxa

State highways transfers .05 (.00, .87)State welfare transfers .44 (.32, .57)State education transfers no maxa

State other transfers .84 (.23, 1.00)

Source: Confidence intervals are computed using normal approximations and asymptoticstandard errors obtained by the delta method from the asymptotic covariance matrix of thecoefficient estimates of the targeting polynomials.a The polynomial does not have any local maximum values in (0, 1).


30/33

Table 3: Point Estimates and 95% Confidence Intervals of Support Values in the OpenInterval (0,1) That Maximize Pre-midterm to Post-midterm Changes in Local Federal Ex-penditures

Institutionally less complex LFEsMaximum is a voter-oriented targeting value Maximum is not in the voter-oriented range

military employment .39 (.39, .39) civilian employment .98 (.00, 1.00)

civilian procurements .33 (.32, .35)civilian salaries .41 (.41, .41)military procurements .42 (.34, .51)military salaries .39 (.26, .53)transfer payments .38 (.14, .67)direct payments .32 (.27, .38)

Institutionally complex LFEsMaximum is a voter-oriented targeting value Maximum is not in the voter-oriented range

Federal welfare transfers .46 (.42, .51) Federal other transfers no maxa

Federal highways transfers .35 (.00, 1.00) State highways transfers .17 (.07, .33)Federal education transfers .32 (.27, .38) State welfare transfers no maxa

State education transfers no maxa

State other transfers .53 (.49, .56)

Source: Confidence intervals are computed using normal approximations and asymptoticstandard errors obtained by the delta method from the asymptotic covariance matrix of thecoefficient estimates of the targeting polynomials.a The polynomial does not have any local maximum values in (0, 1).


31/33

Figure 1: Effects of Support on Federal Local Expenditures, Pre-midterm

civilian procurements (3)

support before midterm

effect

0.30 0.55 0.80

-0.

7

-0.4

-0.

1

military procurement (4)


effect

0.30 0.55 0.80

-0.

5

-0.

2

0.1

civilian employment (2)


effect

0.30 0.55 0.80

-0.0

3

0.

0

0.

02

military employment (3)


effect

0.30 0.55 0.80

-0.

01

0.0

2

0.

04

civilian salaries (3)


effect

0.30 0.55 0.80

-0.

05

-0.

01

0

.02

military salaries (2)


effect

0.30 0.55 0.80

-0.

15

0.

0

0.1

0

transfer payments (2)


effect

0.30 0.55 0.80

-0.

008

0.

0

0.

00

8

direct payments (4)


effect

0.30 0.55 0.80

-0.

6

0.

0

0.

6

1.

2

Fed highways xfers (3)


e

ffect

0.30 0.55 0.80

-2

0

2

4

6

8

Fed welfare xfers (3)


e

ffect

0.30 0.55 0.80

-0.

7

-0.

3

0.1

0.

4

Fed education xfers (4)


e

ffect

0.30 0.55 0.80

-0.

6

-0.

2

0.

2

0.6

Fed other xfers (2)


e

ffect

0.30 0.55 0.80

-1.2

-0.

6

0.0

0.

4

State highways xfers (3)


effect

0.30 0.55 0.80

-1.

0

-0.

4

0.

2

0.

8

State welfare xfers (2)


effect

0.30 0.55 0.80

-0.5

-0.3

-0.

1

0.1

State education xfers (2)


effect

0.30 0.55 0.80

-0.

16

-0.

06

0.

02

State other xfers (4)


effect

0.30 0.55 0.80

-0.

6

-0.3

0.

0

0.

2

Notes: Quasi-likelihood estimates. The number in parentheses shows the degree of the targeting polynomia


32/33

Figure 2: Effects of Support on Local Federal Expenditures, Post-midterm

civilian procurements (3)

support after midterm

effect

0.1 0.4 0.7

-0.

7

-0.4

-0.

1

military procurement (4)


effect

0.1 0.4 0.7

-0.

5

-0.

2

0.1

civilian employment (2)


effect

0.1 0.4 0.7

-0.0

3

0.

0

0.

02

military employment (3)


effect

0.1 0.4 0.7

-0.

01

0.0

2

0.

04

civilian salaries (3)


effect

0.1 0.4 0.7

-0.

05

-0.

01

0

.02

military salaries (2)


effect

0.1 0.4 0.7

-0.

15

0.

0

0.1

0

transfer payments (2)


effect

0.1 0.4 0.7

-0.

008

0.

0

0.

00

8

direct payments (4)


effect

0.1 0.4 0.7

-0.

6

0.

0

0.

6

1.

2

Fed highways xfers (3)


e

ffect

0.1 0.4 0.7

-2

0

2

4

6

8

Fed welfare xfers (3)


e

ffect

0.1 0.4 0.7

-0.

7

-0.

3

0.1

0.

4

Fed education xfers (4)


e

ffect

0.1 0.4 0.7

-0.

6

-0.

2

0.

2

0.6

Fed other xfers (2)


e

ffect

0.1 0.4 0.7

-1.2

-0.

6

0.0

0.

4

State highways xfers (3)


effect

0.1 0.4 0.7

-1.

0

-0.

4

0.

2

0.

8

State welfare xfers (2)


effect

0.1 0.4 0.7

-0.5

-0.3

-0.

1

0.1

State education xfers (2)


effect

0.1 0.4 0.7

-0.

16

-0.

06

0.

02

State other xfers (4)


effect

0.1 0.4 0.7

-0.

6

-0.3

0.

0

0.

2

Notes: Quasi-likelihood estimates. The number in parentheses shows the degree of the targeting polynomia


33/33

Figure 3: Effects of Support on Pre-midterm to Post-midterm Changes in Local FederalExpenditures

civilian procurements

support

targetedc

hange

0.30 0.55 0.80

-0.

1

0.1

0.

3

0.

5

military procurements

support

targetedc

hange

0.30 0.55 0.80

-0.

40

-0.2

0

0.0

civilian employment

support

targetedc

hange

0.30 0.55 0.80

-0.

025

0.0

0.0

15

military employment

support

targetedc

hange

0.30 0.55 0.80

-0.

06

-0.0

2

0.

01

civilian salaries

support

targeted

change

0.30 0.55 0.80

-0.

05

-0.

01

0.

03

0.0

7

military salaries

support

targeted

change

0.30 0.55 0.80

-0.3

0

-0.

15

0.0

transfer payments

support

targeted

change

0.30 0.55 0.80

-0.

016

-0.

008

0.0

direct payments

support

targeted

change

0.30 0.55 0.80

-0.

05

0.

15

0.

35

Fed highways xfers

support

targetedc

hange

0.30 0.55 0.80

-1

1

2

3

4

5

6

Fed welfare xfers

support

targetedc

hange

0.30 0.55 0.80

-0.4

0.

0

0.

4

0.

8

Fed education xfers

support

targetedc

hange

0.30 0.55 0.80

-0.1

0.

2

0.

5

0.8

Fed other xfers

support

targetedc

hange

0.30 0.55 0.80

-0.

2

0.

2

0.

6

1.

0

State highways xfers

support

targeted

change

0.30 0.55 0.80

-0.

4

0.

0

0.4

0.8

State welfare xfers

support

targeted

change

0.30 0.55 0.80

-0.

10

0.

10

0.3

0

0.5

0

State education xfers

support

targeted

change

0.30 0.55 0.80

0.0

0.

06

0.

12

State other xfers

support

targeted

change

0.30 0.55 0.80

-0.2

0.

1

0.

4

Notes: Computed using the targeting polynomial estimates shown in Figures 1 and 2.

Pork Barrel Presidente

Documents

Transcript of Pork Barrel Presidente